Enantioselective Processes at Surfaces Studied by High-Dimensional Neural Network Potentials

点击翻译按钮获取中文摘要

Studying molecule-surface interactions is crucial for the understanding of many important processes ranging from heterogeneous catalysis to life science. In particular the interaction of chiral organic molecules with solid surfaces is of high relevance in both fields. In recent years enantioselective processes at surfaces have emerged as a promising new tool in heterogeneous catalysis for the production of enantiopure pharmaceuticals. The underlying processes, however, are poorly understood at the atomic level thus hindering systematic progress. In particular, the theoretical investigation of these processes is hampered by the large systems, preventing a direct application of modern computational chemistry tools like density-functional theory (DFT). The aim of the current project is to develop, implement and test a new type of neural network potential for high-dimensional multicomponent systems, which is based on DFT and correlated methods, but is much faster to evaluate. This potential will be applied to a detailed study of the individual steps of enantioselective heterogeneous catalysis under realistic conditions. As it allows structural and dynamical studies of very large systems, the method is general and will be applicable to a wide range of complex chemical reactions.